Prediction of mixing efficiency of pharmaceutical€¦ · Prediction of mixing efficiency of...
Transcript of Prediction of mixing efficiency of pharmaceutical€¦ · Prediction of mixing efficiency of...
~ 328 ~
The Pharma Innovation Journal 2017; 6(8): 328-333
ISSN (E): 2277- 7695
ISSN (P): 2349-8242 NAAS Rating 2017: 5.03 TPI 2017; 6(8): 328-333
© 2017 TPI www.thepharmajournal.com
Received: 22-06-2017
Accepted: 23-07-2017
Preeti
Lloyd Institute of Management
& Technology, Greater Noida,
U.P, Dr. A.P.J. Abdul Kalam
Technical University,
Lucknow, Uttar Pradesh. India
Dr. Vandana Arora
Lloyd Institute of Management
& Technology, Greater Noida,
U.P, Dr. A.P.J. Abdul Kalam
Technical University,
Lucknow, Uttar Pradesh. India
Preeti Maan
Lloyd Institute of Management
& Technology, Greater Noida,
U.P, Dr. A.P.J. Abdul Kalam
Technical University,
Lucknow, Uttar Pradesh. India
Correspondence
Preeti
Lloyd Institute of Management
& Technology, Greater Noida,
U.P, Dr. A.P.J. Abdul Kalam
Technical University,
Lucknow, Uttar Pradesh. India
Prediction of mixing efficiency of pharmaceutical
granules
Preeti, Dr. Vandana Arora and Preeti Maan
Abstract Mixing of ingredients, or dispersion of one phase into another, is an operation widely and most
commonly used in pharmaceutical Industry. It is difficult to find a pharmaceutical product in which
mixing is not done at one stage or the other during its manufacturing. Whenever a product contains more
than one component a mixing or blending stage is must required in the manufacturing process to ensure
even distribution of active ingredients, to ensure an even appearance, to ensure that he dosage form
releases the drug at the correct dose at the desired rate etcA pharmaceutical formulation with low amount
of active ingredient will be in effective or with too high amount of active ingredient may be lethal. So,
mixing is very important process and the final mixture required should be completely mixed and
homogenous for such type of formulations
Keywords: Prediction, pharmaceutical granules, homogenous
Introduction
Mixing is defined as the intermingling of two or more dissimilar portions of materials resulting
in attainment of a desired level of uniformity either physically or chemically or as the process
in which two or more than two components in a separate or roughly mixed condition are
treated in such a way so that each particle of any one ingredient lies as nearly as possible to the
adjacent particles of other ingredients or components. The most important use of mixing is the
production of a homogeneous blend of several ingredients which neutralizes variations in
concentration using the minimum amount of energy and time.
Mechanisms of solid-solid mixing Mixing is defined as the intermingling of two or more dissimilar portions of materials resulting
in attainment of a desired level of uniformity either physically or chemically or as the process
in which two or more than two components in a separate or roughly mixed condition are
treated in such a way so that each particle of any one ingredient lies as nearly as possible to the
adjacent particles of other ingredients or components. The most important use of mixing is the
production of a homogeneous blend of several ingredients which neutralizes variations in
concentration using the minimum amount of energy and time.
Mechanisms of solid-solid mixing The homogeneity of the solid-solid mixing and mixing quality is determined pharmaceutically
by mixing time. The mechanism of mixing process of solids can be of dispersive or convective.
Dispersive mixing is mainly intended for the completely random change of place of the
individual particles. If the ingredients are spatially separated at the beginning of the process,
long times will be required to mix them through dispersive mixing alone, since there is a very
low number of assorted neighbors.
Convective mixing causes a movement of large groups of particles relative to each other and is
also a sort of macro mixing. Convection increases the number of assorted neighbors and
thereby promotes the exchange processes of mixing. A material mass is divided up or
convectively mixed through the rearrangement of a solid‟s layers or by the fall of a stream of
material on another layer. The whole volume of material is continuously divided up into
groups and then mixed again. The forced convection can be achieved by rotating element
paddles. The dimension of the groups, which are composed of just one unmixed ingredient, is
continuously reduced due to splitting action of the rotating paddles.
~ 329 ~
The Pharma Innovation Journal
Prediction of mixing efficiency
The variance calculated by using Lacey index (b) is much
easier and simplified to calculate or processed as compared to
other types of variation discussed above. So, that we used the
lacey index equation to determinethe mixing efficiency in our
experimentation work.
In manual procedures, physical sampling of mixture is done
while keeping in mind the 3D aspects of mixture. No doubt,
the maximum accuracy in predicting mixing efficiency can be
obtained by these physical methods, but still these are very
laborious and time consuming procedures. The speed of
prediction of mixing efficiency can be much more increased
by switching from 3D sampling to 2D sampling and that too
in a non-invasive manner. The digital images are among such
2D samples, which can allow mixing efficiency calculations
from surface analysis. The mixing efficiency is predicted in
physical sampling by number fractions or weight fractions.
Similar type of fractions and then variances can be calculated
from 2D surface images also. It is aim of current study to find
out whether the same purpose can be fulfilled by the 2D
surface images or not. Further, in physical sampling, samples
withdrawal manually requires stopping of mixer and causes
disturbance to mixture, sometimes it may also require
complete discharge of mixture from the mixer. These all
difficulties with the 3D sampling can be avoided by using the
2D images. The analysis from images is known as image
analysis and it can be completely automated for mixing
processes and does not causes any disturbance to the mixture.
In addition, image analysis methods offer several advantages
viz storage of data for future access, equal applicability in
small and large processing lots, options for real time studies,
and generating feedback loop system for automatic control of
mixing process. Due to the advantages we have planned to
study image analysis method to predict the mixing efficiency
of solid particles against standard physical sampling method.
Advantage of Image Analysis Image analysis methods offer several advantages viz. storage
of data for future access, equal applicability in small and large
processing lots, options for real time studies, and generating
feedback loop system for automatic control of mixing
process. Due to the advantages we have planned to study
image analysis method to predict the mixing efficiency of
solid particles against Standard physical sampling method.
The images will be taken from the surface of the binary mixer
by a digital camera at various intervals Instead of manual
sampling and will be processed and analysed. It is being
expected that the 2D images can serve the same purpose as
that of manual sampling without any disturbances to the
mixture. For image analysis the particles used for mixing
should be of different colors but should be differentiable, e.g.
color particles. The objective of present study is to calculate
the mixing efficiency by both image analysis and physical
sampling and then comparing them for equivalence studies on
the basis of correlation/similarity.
Work Envisaged
Rational of study
The main objective of this proposed work was to study
mixing efficiency of the mixture of two different colored
pharmaceutical granules by image analysis and physical
counting methods and then to correlate the results of both
methods to examine the usefulness of image analysis method
of predicting mixing efficiency.
Objectives of work
To find out the mixing efficiency of solid/solid mixture by
physical sampling and image analysis method Correlation and
comparison of manual and image analysis technique.
Plan of work
The following plan of work was proposed to achieve desired
goals/objectives.
1. Literature survey
2. Selection of particle material
a) Selection of spherical particles
b) Selection of non-spherical particles
3. Mixing experiment
(a) Mixing experiment by physical sampling
(b) Mixing experiment by image analysis
(c) Mixing experiment by mixed procedure
4. Comparison of methods
(a) Correlation analysis
(b) Optimum mixing time
~ 330 ~
The Pharma Innovation Journal
(c) Error analysis
(d) Similarity test (f2)
Materials
Black/yellow color mustard seeds were purchased from local
market, Rohtak. Precipitated calcium carbonate was
purchased from Avantor Performance Material Ltd., India.
Shellac flakes, ethyl alcohol were purchased from CDH, New
Delhi, India. Ethyl alcohol used was of analytical grade. All
chemicals used were at least lab grade
Equipments
Sigma mixer Make by Shakti Pharma Tech., Ahmadabad, India
Blender Rolex Double Cone Blender
Sieves Set of sieves from sieve no. 22-30 of BSP
Weighing
balance Analytical weighing balance
Digital
camera
SLR camera fitted with 18-55 mm zoom lens
(make by Sony Nex-5, Thailand)
Selection of mixing particles
As the image analysis method was studied for determining the
mixing efficiency of millimeter sized particles; both spherical
as well as non-spherical particles were considered for mixing
experiments in 1:1 wt% proportion.
Selection of spherical particles
Mustard seeds of two different colors (yellow and brown)
were used as spherical particles because mustard seeds are of
equal and uniform size and can be differentiated from each
other easily by color.
Selection of non- spherical particles
For non-spherical particles, pharmaceutical granules of
calcium carbonate (white and blue) were prepared by wet
granulation method using 30%w/v ethanolic solution of
shellac as granulating agent. The wet mass was passed
through a set of 22-30 BSS sieves and dried in air for 8h. The
dried pharmaceutical granules were further passed through
same sieve set in order to get double refines 22-30 sieve size
granules for mixing experiment.
Mixing Experiments
Mixing experiment by physical sampling
The mixing experiment was carried out in a R&D scale
sigma-mixer (Make: Shakti Pharma Tech, model no. SHMD-
AC, CE) fitted with camera at a fixed length for capturing
images. At time𝑡 = 0𝑠, equal wt. of both yellow and brown
color seeds were placed in two layers depicting total unmixed
state. The mixer was started and operated at a speed of 45 rpm
and sampling was done physically at fixed time intervals of
20s up to 300s and then at every 300s up to 3000s. These
spots were uniformly distributed throughout the mixing
surface, such that adjacent sampling points were not affected
by sampling and were at maximum distance from each other.
After sampling black and yellow particles were counted in
each sample and there fractions were calculated for
calculating further variance values "𝜎𝑟2". The "𝜎0 2" was
calculated from 𝑡 = 0𝑠 readings. After calculating variances,
mixing index "𝑏" was calculated. The experiment was
performed five times in order to report error values. The
experiment was repeated under same conditions for non-
spherical (pharmaceutical granules) particles except for
different sampling time intervals, i.e. 2s up to 20s, then 4s up
to 60s, followed by 8s up to 100s and 12s up to 160s and at
last 16s up to 208s.
Sampling points in the Sigma-arm mixer
Mixing experiment by image analysis
As the mixer was fitted with SLR camera, so images were
also captured at all sampling time points just before physical
sampling. Thus for the mixing analysis, sampling was done
by images at same time points used for physical sampling.
The camera was pre setted at; focal length of 4.5mm, shutter
speed of 1/8s and at a fixed distance of 1.5 feet from mixture
surface. The resolution of each digital image was 4592×3056
pixels. After taking images, images were processed through
various steps to finally calculate the value of mixing index
“bIA”. Steps adopted for calculating mixing index were:
Image processing and “Mixing index” calculations
Image J 1.34S software of the National Institute of Health,
USA was used for processing of images of mixture [5]. On
each image, sampling spots as shown in Fig.3 were selected
and then converted into binary spot images. The black and
white area fractions were calculated using “set measurement”
command of software under “Analyze tab”. The "𝜎𝑟2" and
"𝜎02" were calculated using area fractions followed by
calculation of “bIA”. The results were reported as average
“bIA” values vs. time. The flow chart of procedure for digital
image processing for area fraction measurements is given
below
Mixing experiment by mixed method
In addition to physical sampling and image analysis,
intermediate method involving manual counting of different
~ 331 ~
The Pharma Innovation Journal
particles from digital images was also carried out. Mixing
index “bMM” was calculated on the basis of numbers of
particles counted from image manually. This also resulted
into a number based Mixing index “bMM” and the results
were reported as “bMM” value vs. time.
Comparison of methods
Correlation analysis
Various “b” values calculated above by physical sampling, mixed
method and image analysis were correlated with each other for
explaining the usefulness of techniques. Among these methods physical
sampling was the reference method and image analysis was the test
method. The correlation coefficients (r) were calculated for the “bi”
values obtained for all the three methods. A value of correlation
coefficient greater than 0.9 was considered as a highest level of
correlation between the methods.
Similarity test (f2)
A similarity test f2 was applied for comparison of data sets obtained for
various methods. The f2 value is a point-to-point comparison of values at
common time points. The method is based on calculation of residual
values and gives quantitative comparison instead of qualitative or
graphical comparison. The general equation for f2 value is represented as
follows:
Where n is no. of samples, r is reference “bPS” value and t are test “bi”
values. The f2 value˃50 describes the similarity of two data sets while
the value˂50 indicates dissimilarity between data sets.
Optimum mixing time
The optimum mixing times were predicted by tangents intersection
point in “bi” vs. time graphs for both spherical and non-spherical
particles. The intersection point described the time, where mixing is
maximum or variation is minimum. So, comparison of optimum
mixing times from various methods acted as a parameter for setting
usefulness or non-usefulness of test methods.
Error analysis
The %CV was calculated from repeated experiments using formula
given as:
Where SD = Standard Deviation.
Result & Discussion
Mixing by physical sampling
The data set of mixing index (bPS) for five independent
mixing experiments involving spherical and non spherical
particles have been given in Table 5.1 and 5.2. The value of
mixing index "𝑏" are ˃1 for spherical particles and ˂ 1 for
granules or non-spherical (granules) particles. This variation
in standard deviation (SD) (Fig. 5.1 & 5.2) for spherical
particles is almost half as that for non-spherical particles. The
difference may be spotted due to the lackness of sphericity in
granules as compared to spherical particles, which might have
created difficulty in mixing in case of non-spherical particles.
Mixing by mixed method The data set of mixing index “bMM” for five independent
mixing experiments involving spherical and nonspherical
particles have been given in Table 5.4 and 5.5. The value of
mixing index “bMM” are ˃1 for spherical particles and ˂ 1
for non-spherical (pharmaceutical granules) particles.. The
difference in SD is not so large, hence, both spherical and
non-spherical mixings are appearing similarly predicted by
mixedmethod.
Mixing by image analysis
The data sets of mixing index “bIA” for five independent
mixing experiments involving spherical and nonspherical
particles have been given in Table 5.5 and 5.6. The values of
mixing index “bIA” are ˃1 for both spherical and non-
spherical particles. The SD values (Fig. 5.7 & 5.8) are
somewhat larger for non-spherical particles than spherical
particles, however the difference can be considered small as
compared to original values at this high significant figures
level.
Comparison of methods
Correlation analysis The correlation coefficients “r” between various methods for
spherical particles have been given in Table 5.7. The “r” was
found to be 0.5 between image analysis and mixed method
while the “r” value was >0.9 for both physical sampling vs
image analysis and physical sampling vs mixed method. This
shows that standard physical sampling method and image
analysis method are equivalent with each other For non-
spherical, correlation coefficient “r” was found to be >0.9
between all methods indicating good correlation between all
methods. On an overall, the value of “r” was greater for the
image analysis method as compared to mixed method, so
image analysis is more similar/equivalent to standard method
and can be used for predicting mixing efficiencies.
Optimum mixing time
Mixing time for spherical particles
The values of optimum mixing time for spherical particles
was found to be about 70s, 60s & 70s by physical sampling,
~ 332 ~
The Pharma Innovation Journal
mixed method and image analysis method. The mixing time is
quite similar for standard physical method and image analysis
method.
Optimum mixing time by mixed method for spherical particles
Mixing time for non-spherical particles The values of optimum mixing time for spherical particles
was found to be about 24s, 12s & 15s by physical sampling,
mixed method and image analysis method (Fig. 5.12, 5.13 &
5.14) respectively. The mixing time predicted by image
analysis and mixed method is appearing half as that predicted
by physical method but still optimum mixing time by image
analysis is more near to that predicted by standard physical
sampling. Therefore, image analysis is again better method,
which can gives result near to standard method.
Error analysis
The graphical plot of %CV for all the three methods have
been shown in Fig. 5.15 and 5.16 for spherical and non-
spherical particles respectively. The variation in %CV of
image analysis method is least among all for both spherical
and non-spherical particles. This indicated stability of image
analysis towards predicting mixing efficiency. Further the
profile of %CV of image analysis is matching with physical
sampling indicating similarity of image analysis method with
standard method. The variations in mixed method were
deviating by large, so method was not considered useful.
Fig. 5.3 % Coefficient of variation (%CV) of mixing indices for
spherical particles by various methods
Fig 5.4 % Coefficient of variation (%CV) of mixing indices for non-
spherical particles by various methods
Similarity test (f2) analysis
The similarity factors (f2) between various methods for
spherical and non-spherical particles have been given in Table
5.9 & 5.10. The f2 value were found to be >50 for all
methods. This shows similarity between all methods. On an
overall, the “f2” value was greater for the image analysis
method as compared to mixed method for both spherical and
non-spherical particles, so image analysis is maximum useful
for predicting mixing efficiencies.
Summary & Conclusion
In this work, the mixing efficiency of solid-solid mixing of
two different color particles by means of the mixing index at
different time intervals was calculated. For this particular
purpose, Mixing index was calculated by using different
methods i.e. physical sampling, mixed method and image
analysis method. All the methods used for predicting mixing
efficiency was compared and correlated with each other.
The two types of particles; spherical and non-spherical
particles were taken for mixing experiment. Mustard seeds
(yellow and brown color) were selected as spherical particles
and pharmaceutical granules (white and blue color) were used
as non-spherical particles. Methylene blue dye was blended
with calcium carbonate blue color pharmaceutical granules.
The mixing experiment was firstly done by spherical particles
in a sigma-arm mixer. The speed of mixer was fixed at 45
rpm and sampling was done at prefixed time points. The
samples were taken from five different locations within the
mixer. The experiment was repeated five times for average
value. The above experiment was also performed separately
for non-spherical particles in a similar manner as that for
spherical particles.
The no. based fraction of particles were calculated for
physical samples and mixed methods. The area fraction of
particles was measured with the help of Image J software for
image analysis method. From the above fractions, the mixing
~ 333 ~
The Pharma Innovation Journal
index “bi” were calculated and used for relating different
methods.
All the methods used for mixing efficiency prediction were
compared with each other by means of Correlation analysis,
optimum mixing time calculation, error analysis and
similarity test applied on “bi” values. For spherical particles,
the value of correlation coefficient “r” was >0.9 for physical
sampling vs image analysis and physical sampling vs mixed
method and 0.5 for image analysis and mixed method. For
non-spherical particles, the value of correlation coefficient “r”
was found to be >9 between all the methods.
The highest values of “r”were found for physical sampling vs
image analysis. Hence, image analysis can be used instead of
mixed method.
The mixing time calculated from tangent intersection points
showed that in case of spherical particles, as well as non-
spherical particles, image analysis was the For spherical and
non-spherical particles, the %CV was found to be least for
image analysis method. The variations in mixed method were
deviating by large, so this method was not recommended.
The values of similarity factor “f2” between all the methods
were found to be > 50 for both spherical and non-spherical
particles which showed the similarity between methods. The
image analysis method showed highest similarity with
standard method (physical sampling), hence finished for
considering as replacement for standard physical mixing
method.
It is concluded that the image analysis method is similar to
standard physical sampling method in predicting mixing
efficiency. So, we can use the image analysis in place of
physical sampling method. The use of image analysis in place
of physical sampling makes the experiment very simple and
easy and non-invasive.
In addition to this, image analysis having so many other
advantages over physical sampling makes itself a best
method.
Acknowledgements
During my course of work to fulfil the 127 requirements of
this project, I have been 128 accompanied and supported by
many people. I 129 am sincerely thankful to each and every
person 130 who extended their support to me in terms of 131
their knowledge that has been a guiding source 132 for me. I
am highly grateful to my 133 supervisor Dr. Vandana Arora,
Lecturer, 134 Lloyd Institute of Management and 135
Technology, Greater Noida, for her continued 136 support
and valuable suggestions throughout 137 my M.Pharm
education, especially during my 138 project that enabled me
to complete my thesis. 139 It is my privilege to express
humble regards 140 to Jubilant Generics Ltd., R&D Centre,
141 Noida, whose expertise helped me to carry out 142 my
project successfully.
References
1. Fundamentals of particle technology. Solid-Solid Mixing.
Midland information and publishing technology, 2002,
123.
2. Daumann B, Nirschl H. Assessment of mixing efficiency
if solid mixture by means of image analysis. Pow. Tech,
2008; 188:415-523.
3. Aulton ME. The science of dosage form design. Mixing.
2nd edition, 182.
4. Siiria S, Ylirussi J. Determining for a value of mixing.
Powder technology. Pow. Tech, 2009; 196:309317.
5. Coulson JM, Bauman I, Curic D, Boban M. Mixing of
Solids in different mixing devices. Sadhana, 2008;
33(6):721-731.
6. Bhatt B, Aggarwal SS. Revised mixing, 2007.
7. Badger WL, Banchero JL. Introduction to chemical
engineering. 25th reprint. Mixing, 2010, 605.
8. McCabe WL, Smith JC, Harriot P. Unit operations of
chemical engineering. 5th edition, 1993, 942.
9. Holdich RG, Richardson JF. Particulate solids. Chem.
eng. Fifth edition, 2002, 30-32.
10. Sommer K. Sampling of powders and bulk materials.
Springer Verlage Berlin Heidelburg, Newyork (US)
Tokyo, 1989.
11. Cave RA, Cook JP, Connon CJ, Khutoryanskiy VV. A
flow system for the on-line quantitative measurement of
the retention of dosage forms on biological surfaces using
spectroscopy and image analysis. Int. J. of
Pharmaceutics, 2012; 428:96-102.
12. Koller DM, Posch A, Horl G, Voura C, Radl S, Urbanetz
N, Fraser SD, Tritthart W, Reiter F, Schlingmann M,
Khinast JG et al. Continuous quantitative monitoring of
powder mixing dynamics by nearinfrared spectroscopy.
Pow. Tech, 2011, 205:87-96.
13. Simon LL, Oucherif KA, Nagy ZK, Hungerbuhler K.
Bulk video imaging based multivariate image analysis,
process control chart and acoustic signal assisted
nucleation detection. Chem. eng. Sci, 2010; 65:4983-
4995.
14. Hajra SK, Bhattacharya T. Improvement of granular
mixing of dissimilar materials in rotating cylinders. Pow.
Tech, 2010, 198:175-182.
15. Siiria S, Yliruusi J. Determine a value for mixing: Mixing
degree. Pow. Tech, 2009; 196:309-317.
16. Daumann B, Fath A, Amlauf H, Nirschl H.
Determination of the mixing time in a discontinuous
powder mixer by using image analysis. Chem. eng. Sci,
2009; 64:2320-2331.
17. Derksen JJ. Mixing by solid particles. Chem. eng. res.
Desi, 2008; 86:1363-1368.
18. Baumann I. Solid-solid mixing with static mixer. Bio
chem eng. 2008; 15:159-165.
19. Berthiaux H, Mosorov V, Tomczak L, Gatumel C,
Demeyre JF. Principal component analysis for
characterizing homogeneity in powder mixing using
image processing techniques. Chem. eng. Proc, 2006;
45:397-403.
20. Santomaso A, Olivi M, Canu P. Mechanisms of mixing
of granular materials in drum mixers under rolling
regime. Chem. Eng. Sci, 2004; 59:3269-3280.
21. Kopriva I, Du Q, Szu H, Wasylkiwskyj W. Independent
component analysis approach to image sharpening in the
presence of atmospheric turbulence. Optics Commu,
2004; 233:7-14.